ECS289A, Spring 2004,  Gene Network Inference

Projects List

• Weighted Perturbation Graph (compared to the unweighted in Wagner, 2002)
• Single + Double Perturbations (single and double knockout data is available for yeast, what can we do with it?)
  
• Other Parsimony Arguments (can you come up with a simple yet biologically plausable objective graph scoring function?)
  
• Network Expansion from a Core Using Perturbation Data (given a subnetwork can you expand it using perturbation data?)
  

• Implementing/Modifying a BN learner (some code is available, can we make it better?)
• Space pruning for more efficient local search (Recently more efficient local search algorithms have been proposed. Can we do better?)
• Mean Field Algorithm for learning (Newer Bayes Net learning algorithm)
• Incorporate modularity as a hybrid local probability distribution (between linear and multinomial)

• Implementing a Boolean Network learner, a la Ideker et al
• Robust Boolean Networks
• Restricting Boolean Functions (. ^ . ^ .) v (. ^ . ^ .) ...
• Dimension reduction in Boolean Nets
• Post-classes and Boolean Learners
• Network extension under the Boolean Formalism
• Incomplete-data insensitive Boolean Networks
• Modularizing Boolean Networks

• Data Consistency of Known Networks
• Finding Modules in Existing Networks